ISO 17776 MAH Bow-Tie and EI AVIFF Flow-Induced Vibration Screening

Class Purpose Standard
MahCatalogue Pre-defined threats, consequences and barriers per MAH type ISO 17776
MahBowTieBuilder Assemble a BowTieModel for a major-accident hazard ISO 17776
PipingFivScreening Factored likelihood-of-failure screening for flow-induced vibration Energy Institute AVIFF
FivLikelihoodResult LOF score and likelihood band for one circuit Energy Institute AVIFF
FlowInducedVibrationAnalyser Rigorous LOF from a solved PipeBeggsAndBrills segment Energy Institute AVIFF
FlowInducedPulsationScreening Acoustic lock-in screening for closed side branches (dead legs) Energy Institute AVIFF T2.6
FlowInducedPulsationResult Branch modes, lock-in envelopes and resonance velocities Energy Institute AVIFF T2.6

Classes live under neqsim.process.safety.hazid, neqsim.process.safety.vibration and neqsim.process.measurementdevice.

MAH bow-tie from the ISO 17776 catalogue

MahBowTieBuilder.build(MahType) returns a fully populated BowTieModel (threats on the left, consequences on the right, barriers in the middle) for a standard major-accident-hazard type:

import neqsim.process.safety.hazid.MahType;
import neqsim.process.safety.hazid.MahBowTieBuilder;
import neqsim.process.safety.hazid.MahCatalogue;
import neqsim.process.safety.risk.bowtie.BowTieModel;

BowTieModel bowtie = MahBowTieBuilder.build(MahType.TOPSIDE_HYDROCARBON_RELEASE);

String hazard = bowtie.getHazardId();
bowtie.getThreats();        // ≥ 4 threats, each with getFrequency()
bowtie.getConsequences();   // ≥ 3 consequences
bowtie.getBarriers();       // ≥ 5 barriers, each with getPfd()

// Inspect the raw catalogue entries directly
MahCatalogue.threatsFor(MahType.TOPSIDE_HYDROCARBON_RELEASE);
MahCatalogue.consequencesFor(MahType.TOPSIDE_HYDROCARBON_RELEASE);
MahCatalogue.barriersFor(MahType.TOPSIDE_HYDROCARBON_RELEASE);

Default threat frequency and barrier PFD are exposed as MahBowTieBuilder.DEFAULT_THREAT_FREQUENCY and MahBowTieBuilder.DEFAULT_BARRIER_PFD. MahType covers TOPSIDE_HYDROCARBON_RELEASE, RISER_LEAK, WELL_BLOWOUT, STRUCTURAL_COLLAPSE, DROPPED_OBJECT, HELICOPTER_LOSS, SHIP_COLLISION, FIRE_EXPLOSION, TOXIC_RELEASE, LOSS_OF_BUOYANCY, and EXTREME_WEATHER, each carrying a human-readable description.

EI AVIFF flow-induced-vibration screening

PipingFivScreening computes an Energy Institute AVIFF likelihood-of-failure (LOF) score for a piping circuit and maps it to a likelihood band. Use screenGas or screenLiquid depending on the fluid:

import neqsim.process.safety.vibration.PipingFivScreening;
import neqsim.process.safety.vibration.PipingFivLikelihood;
import neqsim.process.safety.vibration.FivLikelihoodResult;

// Gas circuit: tag, rho[kg/m3], v[m/s], D[m], wall t[m], nBranches, pulsation, support
FivLikelihoodResult gas = PipingFivScreening.screenGas(
    "Compressor discharge", 80.0, 30.0, 0.3, 0.006, 2, 4.0, 2.0);

double lof = gas.getLofScore();
PipingFivLikelihood band = gas.getLikelihood();   // LOW / MEDIUM / HIGH / VERY_HIGH
String json = gas.toJson();                        // contains "lofScore", "likelihood"

// Liquid circuit: tag, v[m/s], D[m], wall t[m], nBranches, support
FivLikelihoodResult liquid = PipingFivScreening.screenLiquid(
    "Pump discharge", 3.5, 0.15, 0.005, 1, 1.5);

// Map an arbitrary LOF score to a band
PipingFivLikelihood b = PipingFivScreening.bandFor(0.7);   // HIGH

The likelihood bands are LOW (< 0.3), MEDIUM (0.3–0.5), HIGH (0.5–1.0), and VERY_HIGH (≥ 1.0). Invalid geometry (zero diameter, negative velocity) throws IllegalArgumentException.

Choosing the right vibration tool

NeqSim has three complementary vibration screenings. They answer different questions and a “pass” on one does not clear the others:

Class Mechanism Question it answers
PipingFivScreening Main-line FIV, factored Quick desktop LOF from density, velocity and D/t
FlowInducedVibrationAnalyser Main-line FIV, rigorous LOF from a solved PipeBeggsAndBrills segment, with the real mixture density, velocity and void fraction
FlowInducedPulsationScreening Tonal FIP at closed side branches Whether a dead leg can lock into acoustic resonance
AcousticInducedVibrationScreening Broadband AIV Sound power downstream of a pressure-reducing device

Rigorous main-line LOF from a solved pipe

FlowInducedVibrationAnalyser evaluates the AVIFF form

\[\mathrm{LOF} = \frac{\rho_m v_m^2\, F_{VF}}{F_v}, \qquad F_v = \alpha\left(\frac{D}{t}\right)^{\beta}\]

on a segment of a PipeBeggsAndBrills that has been run, so the mixture density, mixture velocity and void fraction come from the flow solution rather than from hand estimates.

The fluid-viscosity factor F_VF switches on the void fraction β:

Void fraction β F_VF
β < 0.2 0.2 + 4β
0.2 ≤ β ≤ 0.88 1.0 (liquid and general multiphase)
0.88 < β ≤ 0.99 -27.882 β² + 45.545 β - 17.495 (wet gas)
β > 0.99 sqrt(μ_gas / 1 cP) (gas dominated)

Physical sanity check. F_VF must fall as β goes to 1. The wet-gas branch reaches 0.268 at β = 0.99, so a single-phase gas must come out below that — a hydrocarbon gas at 0.012–0.018 cP gives about 0.11. Removing liquid from a wet-gas line lowers the vibration driver; at equal standard rate and pressure the wet-over-dry driver ratio for a rich gas at 40–50 bara is roughly 3 to 4. If a calculation reports the opposite, F_VF is being evaluated wrongly. REFERENCE_VISCOSITY_CP is exposed as a public constant so the branch can be reproduced independently. Note that PipeBeggsAndBrills.getSegmentMixtureViscosity returns centipoise, not Pa·s.

import neqsim.process.equipment.pipeline.PipeBeggsAndBrills;
import neqsim.process.measurementdevice.FlowInducedVibrationAnalyser;

PipeBeggsAndBrills pipe = new PipeBeggsAndBrills("inlet pipe", feed);
pipe.setDiameter(0.3652);
pipe.setThickness(0.0206);      // REQUIRED: the LOF correlation divides by D/t
pipe.setLength(12.0);
pipe.setNumberOfIncrements(4);

FlowInducedVibrationAnalyser fiv = new FlowInducedVibrationAnalyser("LOF", pipe);
fiv.setMethod("LOF");
fiv.setSupportArrangement("Medium stiff");   // Stiff / Medium stiff / Medium / Flexible

process.add(feed);
process.add(pipe);
process.run();

double lof = fiv.getMeasuredValue("");

Omitting setThickness throws IllegalStateException rather than silently returning NaN. The support arrangement is a qualitative stiffness category, not a support spacing.

Calibrated LOF ratios when the line size is unknown

A common situation is that a design LOF is quoted but the line list is not available. For two operating points on the same line the pipe factor F_v and the flow area cancel exactly, so

\[\frac{\mathrm{LOF}_2}{\mathrm{LOF}_1} = \frac{(\rho_m v_m^2 F_{VF})_2}{(\rho_m v_m^2 F_{VF})_1}\]

is independent of diameter, wall thickness and support category. Reproduce the stated design point with an assumed geometry, then report every other case as a ratio to it. Confirm the cancellation numerically by re-running one case with a different setSupportArrangement(...) — the calibrated ratio must not move.

Flow-induced pulsation at closed side branches

FlowInducedPulsationScreening covers the tonal, acoustically resonant mechanism that main-line FIV screening does not. Flow past the mouth of a dead leg sheds a shear layer; when the shedding frequency falls within ±20 % of a standing acoustic mode of the branch the two lock in and the branch self-excites.

The procedure is:

  1. Acoustic length L — centreline distance from the tee to the first acoustic boundary (normally closed valve, blind, or a large volume such as a separator, cooler or KO drum). No end correction is applied.
  2. Eigenfrequencies — f_n = (2n+1)c/(4L) for a CLOSED termination (R = +1), f_n = (n+1)c/(2L) for OPEN (R = −1), with n starting at 0.
  3. Excitation — f_s = Sr·U0/W_eff where W_eff = π·d_s/4 + r_eff is the effective width of the branch mouth, not the branch diameter. Sr = 0.37 is the recommended screening value for side-branch modes, 0.20 for the main-header mode.
  4. Resonance check — possible when 0.8 f_n ≤ f_s ≤ 1.2 f_n.
import neqsim.process.safety.vibration.FlowInducedPulsationScreening;
import neqsim.process.safety.vibration.FlowInducedPulsationResult;

// name, acoustic length [m], branch ID [m], run ID [m], U0 [m/s], rho [kg/m3], c [m/s]
FlowInducedPulsationResult res = FlowInducedPulsationScreening.screen(
    "Closed cross-over stub", 3.0, 0.2477, 0.3652, 20.1, 46.2, 374.0);

boolean resonance = res.isAnyModeLockedIn();
double fs = res.getSheddingFrequencyHz();
for (FlowInducedPulsationResult.BranchMode m : res.getModes()) {
  // m.getModeIndex(), m.getFrequencyHz(), m.getEnvelopeLowHz(), m.getEnvelopeHighHz(),
  // m.isLockedIn(), m.getResonanceVelocityMPerS()
}

// Helpers for building length or velocity windows without running a full screening
double weff = FlowInducedPulsationScreening.effectiveWidth(0.2477, 0.0);
double f0 = FlowInducedPulsationScreening.eigenFrequency(
    0, 374.0, 3.0, FlowInducedPulsationScreening.AcousticTermination.CLOSED);

The full overload takes the edge radius, the termination, the Strouhal number and the mode count.

Why a wet-gas measurement campaign cannot clear dry-gas service

Main-line FIV relaxes when a line goes dry, but flow-induced pulsation moves the other way. Even a small amount of a second phase affects not only the vortex shedding but also the acoustic damping and the speed of sound in the branch, so drying the gas removes three protections at once: the liquid that damped the resonator, the slugging that disrupted the shear layer, and — for drains — the liquid filling that set the branch sound speed near 850–1000 m/s instead of ~375 m/s. Run-pipe accelerometers are largely blind to branch pulsation, so a clean main-line vibration record does not clear this mechanism.

Because the ±20 % envelope makes the resonant length windows narrow, per-branch verdicts are highly sensitive to the acoustic length. Where as-built lengths are not available, invert the criterion instead and report the resonant length window per branch size:

\[L \in \left[(1-0.2)\frac{(2n+1)c}{4 f_s},\; (1+0.2)\frac{(2n+1)c}{4 f_s}\right]\]

which a walkdown can check directly with a tape measure.

Verification

./mvnw test -Dtest=MahBowTieBuilderTest,PipingFivScreeningTest
./mvnw test -Dtest=FlowInducedVibrationAnalyserTest,FlowInducedPulsationScreeningTest

FlowInducedPulsationScreeningTest reproduces a published worked example: a 3 m closed branch at c = 400 m/s gives f₀ = 33.3 Hz, f₁ = 100 Hz, f₂ = 166.7 Hz. FlowInducedVibrationAnalyserTest asserts that dry-gas LOF stays below wet-gas LOF at equal standard rate and pressure.